In known voice response systems a dialogue model is used to model a dialogue between a user and the system. Often such a dialogue model comprises states (or nodes) which are notionally connected by edges. Conceptually a user making a call to the visits a state and the system asks the user a question in dependence up on the current state the user is visiting. The user's answer is analysed by the system in order to decide which state the user should visit next, and hence what the next question should be.
However, a problem with such a system is that it is possible for the user to get ‘stuck’ in a particular state, and hence the dialogue becomes repetitive. In the worst case the user terminates the call, at the very least the user is discouraged from using the system again even if they do eventually achieve the task they set out to do.